RIZAP announced on September 3 that an employee mistakenly uploaded customer information from a specific health guidance management system to an external generative AI service used for personal purposes.
OpenAI has announced GPT-6 Astra, a new AI model that outperforms previous models in PC operation, programming, and cybersecurity.
OpenAI has announced that it will provide ChatGPT paid users with daily usage limit reset rights in conjunction with the rollout of the new GPT-6 Astra model. These reset rights will be carry-overable.
Microsoft has released MAI-Transcribe-2, a speech recognition model supporting automatic detection of 60 languages. The company claims the model offers higher processing speeds and accuracy compared to existing models from OpenAI and Google.
NVIDIA has announced that PCs equipped with the Arm-based SoC RTX Spark are scheduled to arrive in October 2026, with companies such as Lenovo and ASUS unveiling specific product models.
The RIZAP Group has apologized after an employee of a subsidiary mistakenly uploaded personal information, including customer names and medical conditions, to a private generative AI service.
At the results presentation for the Physical AI Development Support Program hosted by AWS Japan, participating companies showcased technical initiatives aimed at overcoming data scarcity.
NVIDIA has announced the acquisition of Hugging Face, a distribution platform for AI models and code, for $12.9303 billion.
As AI becomes capable of programming, experts are suggesting how the roles of human engineers must evolve.
Adobe announced that Anil Chakravarthy will assume the role of President and CEO on December 1, with current CEO Shantanu Narayanan transitioning to Executive Chairman.
Anthropic has released v2.1.260, the latest version of Claude Code. This update introduces a diff view in full-screen mode and fixes several bugs related to permission management and AWS Bedrock connectivity.
Zumen AI has announced technology that automatically generates 3D CAD models based on 2D drawings. The tool operates in a web browser and includes a feature to create exploded view drawings from assemblies.
OpenAI announced on September 3 that it has released the new GPT-6 Astra model to select organizations. This is the first model to be certified as reaching the highest safety guideline level, Critical, for its cybersecurity capabilities.
A contribution to IEEE Spectrum highlights the risk that AI replacing junior-level tasks may undermine the apprenticeship-like process through which professional expertise is passed down.
The AI-powered wearable device Watch v will be released on October 1 as a measure against customer harassment, using AI to detect inappropriate remarks and verbal abuse from audio.
Gartner has published its Hype Cycle for Future-Oriented Infrastructure for 2026, illustrating the maturity of technologies that will become critical in the future, including AI-related advancements.
Mitsui-Soko Holdings announced a total investment of 20 billion yen as part of its DX strategy leading up to the fiscal year ending April 2026. The company aims to refresh its IT infrastructure and promote the use of AI in logistics operations.
OpenAI has published a technical report summarizing red teaming results conducted with safety restrictions disabled regarding a hacking incident involving Hugging Face.
AI engineer Raphael Pierre introduced six methods for Retrieval-Augmented Generation (RAG), starting with full-text search and introducing complex mechanisms as needed.
An "AI worm" has been reported where Microsoft Copilot accepts malicious instructions embedded in Word documents as commands and spreads the infection by copying them into newly generated documents.